Gamma-ray bursts (GRBs) show different behaviours and trends in their spectral evolution. One of the methods used to understand the physical origin of these behaviours is to study correlation between the spectral fit parameters. In this work, we used a Bayesian analysis method to fit time-resolved spectra of GRB pulses that were detected by the \textit{Fermi}/GBM during its first 9 years of mission. We studied single pulsed long bursts ($T_{90}\geq2$ s). Among all the parameter correlations, we found that the correlation between the low-energy power-law index $\alpha$ and the energy flux exhibited a systematic behaviour. We presented the properties of the observed characteristics of this behaviour and interpreted it in the context of the photospheric emission model.